import pydata_google_auth
import datetime
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
import warnings
import math

from datetime import timedelta
from pytz import timezone, utc

from scipy import stats
import plotly.express as px
import plotly.offline as pyo

from fpdf import FPDF

params = {'project_id' : 7807, 
          'start_date_kst': '2021-05-01', 
          'end_date_kst' : '2021-06-26'
          ,'start_date_utc': '', 
          'end_date_utc': '', 
          'n': 20}

def getBigQueryData(params):
    query = """
        SELECT
        R.*,
        M.member_nm as worker_name,
        M.login_id as worker_login_id
        FROM (SELECT
                CAST(W.project_id as INT64) as project_id,
                CAST(W.worker_id as INT64) as worker_id,
                CAST(W.first_work_cnt as INT64) as first_work_cnt,
                CAST(W.first_work_time as INT64) as first_work_time,
                CAST(W.first_work_avg_time as FLOAT64) as first_work_avg_time,
                CAST(W.re_work_cnt as INT64) as re_work_cnt,
                CAST(W.re_work_time as INT64) as re_work_time,
                CAST(W.re_work_avg_time as FLOAT64) as re_work_avg_time,
                CAST(W.work_tot_cnt as INT64) as work_tot_cnt,
                IFNULL(C.check_cnt,0) as check_cnt, -- 검수 진행된 작업 수
                IFNULL(C.reject_cnt,0) as reject_cnt, -- 누적 반려 수
                ROUND(IFNULL(SAFE_DIVIDE(IFNULL(C.check_cnt,0),CAST(W.first_work_cnt as INT64)),0) * 100, 2) as check_rate_per,  -- 검수진행률
                ROUND(IFNULL(SAFE_DIVIDE(IFNULL(C.reject_cnt,0),CAST(C.check_cnt as INT64)),0) * 100, 2) as reject_rate_per,  -- 반려율
                iqr_work_cnt,
                iqr_work_avg_time
            FROM
            (
                SELECT
                    project_id,
                    worker_id,
                    sum(if(work_data_seq='1', 1, 0)) as first_work_cnt, -- 최초 작업 수
                    sum(if(work_data_seq='1', diff, 0)) as first_work_time, -- 최초 작업시간 합
                    round(ifnull(SAFE_DIVIDE(sum(if(work_data_seq='1', diff, 0)), sum(if(work_data_seq='1', 1, 0))),0), 2) as first_work_avg_time, -- 평균
                    sum(if(work_data_seq <> '1', 1, 0)) as re_work_cnt, -- 재작업 수
                    sum(if(work_data_seq <> '1', diff, 0)) as re_work_time, -- 재작업 시간 합
                    round(ifnull(SAFE_DIVIDE(sum(if(work_data_seq<>'1', diff, 0)), sum(if(work_data_seq<>'1', 1, 0))),0), 2) as re_work_avg_time, -- 재작업시간 평균
                    count(data_id) as work_tot_cnt,
                    sum(if(work_data_seq='1' and diff >= percentile_25 and diff <= percentile_75, 1, 0)) as iqr_work_cnt, -- 최초 작업 수
                    round(ifnull(SAFE_DIVIDE(sum(if(work_data_seq='1' and diff >= percentile_25 and diff <= percentile_75, diff, 0)), sum(if(work_data_seq='1' and diff >= percentile_25 and diff <= percentile_75, 1, 0))),0), 2) as iqr_work_avg_time -- 평균
                FROM (
                    SELECT
                        datetime(cast(dt as timestamp), 'Asia/Seoul') as dt,
                        project_id,
                        data_id,
                        worker_id,
                        datetime(cast(s_dt as timestamp), 'Asia/Seoul') as s_dt,
                        datetime(cast(e_dt as timestamp), 'Asia/Seoul') as e_dt,
                        DATETIME_DIFF(e_dt, s_dt, SECOND) as diff,
                        work_data_seq,
                        percentile_25,
                        percentile_75
                    FROM(
                        SELECT
                         MAX(dt) as dt
                        ,project_id
                        ,data_id
                        ,worker_id
                        ,MIN(IF(action='work_start' or action='rework_start', dt, null)) as s_dt
                        ,MAX(IF(action='work_end' or action='rework_end', dt, null)) as e_dt
                        ,work_data_seq
                        FROM `crowdworks-platform.crowdworks_event.cw_event_data`
                        WHERE
                            day >= DATE(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
                        AND dt >=DATETIME(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
                        AND day <= DATE(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
                        AND dt <= DATETIME(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
                        AND project_id = '{project_id}'
                        AND worker_id <> ''
                        AND data_id <> ''
                        AND action in ('work_start','work_end', 'rework_start', 'rework_end')
                        GROUP BY project_id,data_id,worker_id,work_data_seq
                    ),
                    (
                      SELECT
                          percentile_25,
                          percentile_75,
                          (percentile_75 - percentile_25) as iqr
                      FROM (
                          SELECT
                              APPROX_QUANTILES(diff, 4)[OFFSET (1)] AS percentile_25,
                              APPROX_QUANTILES(diff, 4)[OFFSET (3)] AS percentile_75
                          FROM (
                              SELECT
                                  datetime(cast(dt as timestamp), 'Asia/Seoul') as dt,
                                  project_id,
                                  data_id,
                                  worker_id,
                                  datetime(cast(s_dt as timestamp), 'Asia/Seoul') as s_dt,
                                  datetime(cast(e_dt as timestamp), 'Asia/Seoul') as e_dt,
                                  DATETIME_DIFF(e_dt, s_dt, SECOND) as diff,
                                  work_data_seq
                              FROM (
                                  SELECT
                                   MAX(dt) as dt
                                  ,project_id
                                  ,data_id
                                  ,worker_id
                                  ,MIN(IF(action='work_start' or action='rework_start', dt, null)) as s_dt
                                  ,MAX(IF(action='work_end' or action='rework_end', dt, null)) as e_dt
                                  ,work_data_seq
                                  FROM `crowdworks-platform.crowdworks_event.cw_event_data`
                                  WHERE
                                      day >= DATE(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
                                  AND dt >=DATETIME(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
                                  AND day <= DATE(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
                                  AND dt <= DATETIME(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
                                  AND project_id = '{project_id}'
                                  AND worker_id <> ''
                                  AND data_id <> ''
                                  AND action in ('work_start','work_end', 'rework_start', 'rework_end')
                                  GROUP BY project_id,data_id,worker_id,work_data_seq
                              )
                              WHERE s_dt is not null AND e_dt is not null AND work_data_seq='1'
                          )
                      )
                    )
                    WHERE s_dt is not null AND e_dt is not null
                )
                GROUP BY 1,2
            ) W
            LEFT JOIN
            (
                SELECT
                    project_id,
                    worker_id,
                    count(data_id) as check_cnt,
                    SUM(reject_cnt) as reject_cnt
                FROM (
                    SELECT
                        project_id,
                        worker_id,
                        data_id,
                        SUM(reject) as reject_cnt
                    FROM (
                        SELECT
                            A.action as action,
                            A.dt as dt,
                            A.project_id as project_id,
                            A.data_id as data_id,
                            A.worker_id as worker_id,
                            A.work_data_seq as work_data_seq,
                            B.rank as rank,
                            if(a.action = 'check_reject', 1, 0) as reject
                        FROM
                            (SELECT
                                action,
                                dt,
                                project_id,
                                data_id,
                                worker_id,
                                work_data_seq,
                            FROM `crowdworks-platform.crowdworks_event.cw_event_data`
                            WHERE
                                    day >= '{start_date_kst}'   AND day <= '{end_date_kst}'
                                AND dt >= '{start_date_kst}' AND dt <= '{end_date_kst}'
                                AND project_id = '{project_id}'
                                AND worker_id <> ''
                                AND data_id <> ''
                                AND checker_id <> ''
                                AND action in ('check_end','check_reject')) A
                            LEFT JOIN
                                (SELECT
                                    action,
                                    dt,
                                    project_id,
                                    data_id,
                                    worker_id,
                                    work_data_seq,
                                    RANK() OVER (PARTITION BY worker_id ORDER BY dt) as rank
                                FROM `crowdworks-platform.crowdworks_event.cw_event_data`
                                WHERE
                                        day >= DATE(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
                                    AND dt >=DATETIME(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
                                    AND day <= DATE(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
                                    AND dt <= DATETIME(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
                                    AND project_id = '{project_id}'
                                    AND worker_id <> ''
                                    AND data_id <> ''
                                    AND action in ('work_end','rework_end')) B
                                ON (A.data_id = B.data_id AND A.worker_id = B.worker_id AND A.work_data_seq = B.work_data_seq)
                    )
                    GROUP BY 1,2,3
                )
                GROUP BY 1,2
            ) C
             ON (W.project_id = C.project_id AND W.worker_id = C.worker_id)
        ) R
        LEFT JOIN `crowdworks-platform.dgwp_1st_party.works_user_profile` M
        ON (M.member_id = R.worker_id)
    """.format(project_id=params['project_id'], start_date_kst=params['start_date_kst'], end_date_kst=params['end_date_kst'])
    
    df = pd.read_gbq(query = query, project_id = 'crowdworks-platform', dialect='standard')
    
    query2 = """
        SELECT
            day,
            dt,
            action,
            worker_id,
            project_id,
            data_id
        FROM `crowdworks-platform.crowdworks_event.cw_event_data`
        WHERE
                day >= DATE(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
            AND dt >=DATETIME(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
            AND day <= DATE(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
            AND dt <= DATETIME(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
        AND project_id = '{project_id}'
        AND worker_id <> ''
        AND data_id <> ''
        ORDER BY dt
    """.format(project_id=params['project_id'], start_date_kst=params['start_date_kst'], end_date_kst=params['end_date_kst'])
    
    df2 = pd.read_gbq(query = query2, project_id = 'crowdworks-platform', dialect='standard')
    
    query3 = """
        SELECT
            project_id,
            data_id,
            sum(if(action = 'check_reject', 1, 0)) as check_reject_cnt,
        FROM `crowdworks-platform.crowdworks_event.cw_event_data`
        WHERE
            day >= DATE(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
        AND dt >=DATETIME(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
        AND day <= DATE(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
        AND dt <= DATETIME(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
        AND project_id = '{project_id}'
        AND worker_id <> ''
        AND data_id <> ''
        AND action in ('check_end','check_reject')
        GROUP BY 1,2
    """.format(project_id=params['project_id'], start_date_kst=params['start_date_kst'], end_date_kst=params['end_date_kst'])
    
    df3 = pd.read_gbq(query = query3, project_id = 'crowdworks-platform', dialect='standard')
        
    query4 = """
        with d as  (SELECT
            project_id,
            member_id,
            worker_id,
            action,
            data_id
        FROM `crowdworks-platform.crowdworks_event.cw_event_data`
        WHERE
            day >= DATE(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
        AND dt >=DATETIME(DATETIME_ADD('{start_date_kst} 00:00:00', INTERVAL -9 HOUR))
        AND day <= DATE(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
        AND dt <= DATETIME(DATETIME_ADD('{end_date_kst} 23:59:59', INTERVAL -9 HOUR))
        AND project_id = '{project_id}'
        AND worker_id <> ''
        AND data_id <> ''
        AND action in ('work_start', 'work_end','all_finished'))
        select 
            project_id,
            MAX(s1) as s1,
            MAX(s2) as s2,
            MAX(s3) as s3
        from
        (
            select
                project_id,
                count(distinct member_id) as s1,
                0 as s2,
                0 as s3
            from d
            where action='work_start'
            group by 1
        union all 
            select
                project_id,
                0 as s1,
                count(distinct member_id) as s2,
                0 as s3
            from d
            where action='work_end'
            group by 1
        union all 
            select 
                project_id,
                0 as s1,
                0 as s2,
                count(distinct worker_id) as s3
            from 
            (select 
                project_id,
                worker_id,
                count(distinct data_id) data_cnt
            from d
            where action='all_finished'
            group by 1,2
            having data_cnt >= {n})
            group by 1
        )
        group by 1
    """.format(project_id=params['project_id'], start_date_kst=params['start_date_kst'], end_date_kst=params['end_date_kst'], n=params['n'])

    df4 = pd.read_gbq(query = query4, project_id = 'crowdworks-platform', dialect='standard')
    
    return df, df2, df3, df4

import pymysql

conn = pymysql.connect(
    host='db-1ea6m.pub-cdb.ntruss.com',
    port=int(3306),
    user='ybkim',
    passwd='m883jPts',
    db='cwaidata'
)

project_info = pd.read_sql_query("""
    SELECT *
    FROM CW_PROJECT
    WHERE project_id = '{project_id}'
    """.format(project_id=params['project_id']), conn)

project_info2 = pd.read_sql_query("""
    SELECT *
    FROM TB_PRJ_MST
    WHERE project_id = '{project_id}'
    """.format(project_id=params['project_id']), conn)

df, df2, df3, df4 = getBigQueryData(params)

1. 프로젝트 Overview

작업자 수 & 작업량 수 daily

목적: 현재까지의 진행 상황

df2[df2['action']=='work_end'].groupby('day').count()['data_id']

day
2021-06-03      457
2021-06-04      755
2021-06-05      190
2021-06-06       65
2021-06-07      666
2021-06-08      212
2021-06-09     1298
2021-06-10      796
2021-06-11    13106
2021-06-12    14275
2021-06-13    10378
2021-06-14     8951
2021-06-19        2
Name: data_id, dtype: int64

fig, axs = plt.subplots(1,2)
fig.subplots_adjust(wspace=.5)
plt.rcParams['figure.figsize'] = [15, 5]

#날짜별로 그룹 하고 워커 아이디의 유니크한 수를 계산
df2.groupby('day')['worker_id'].nunique().plot(kind='line', title='number of workers by day', ax=axs[0])

#action이 work_end인 애들로만 필터를 하고 날짜별로 그룹 한 다음 data 수를 count함
df2[df2['action']=='work_end'].groupby('day').count()['data_id'].plot(kind='line', title='number of tasks completed by day', ax=axs[1])

plt.savefig('overview.jpg')

작업자 Funnel

목적: 작업자 이탈 하는 포인트가 어디인지

용어: enter_num = 작업 화면 페이지 입장자 수. trial_num = 최소 1개의 작업을 한 작업자 수. worker_num = n개의 작업물 이상을 한 작업자 수

df4

	s1	s2	s3
0	1494	929	0

df4 = df4.drop(columns='project_id')

pyo.init_notebook_mode()

new_df4 = pd.DataFrame()
new_df4['number'] = df4.values[0]
new_df4['stage'] = ['enter_num', 'trial_num', 'worker_num']
fig = px.funnel(new_df4, x='number', y='stage')
fig.show()
fig.write_image("funnel.jpg") 

활동적인 작업자 수

최소 n개 이상을 한 작업자 수를 daily로 보는 그래프

data_cnt_by_day_and_worker[data_cnt_by_day_and_worker[0]>params['n']].groupby('day').count()[0]

#work_end로 필터하고, 날짜별, 워커 아이디별 그룹을 하고, 각 날짜의 그 워커가 몇개의 작업을 했는지 카운트 한다
data_cnt_by_day_and_worker = df2[df2['action']=='work_end'].groupby(['day', 'worker_id']).size().reset_index()

#n개 이상을 한 워커들로 필터를 하고, 날짜별로 그룹핑을 하고, 카운트 함
active_worker_count_day = data_cnt_by_day_and_worker[data_cnt_by_day_and_worker[0]>params['n']].groupby('day').count()[0]

title = 'Count of Workers that Submited at Least ' + str(params['n']) + ' tasks'
active_worker_count_day.plot(kind='line', title=title)
plt.ylabel('Active Worker Count')
plt.savefig('active_worker_count.jpg')

활동적인 작업자들의 daily 작업량

average_work_count_per_worker_day = data_cnt_by_day_and_worker[data_cnt_by_day_and_worker[0]>params['n']].groupby('day').mean()[0]
average_work_count_per_worker_day.plot(kind='line', title = 'Daily Average Count of Tasks Submitted by Active Worker')
plt.ylabel('Task Count')
plt.savefig('daily_average_task_count_by_worker.jpg')

2. 작업 난이도

작업 스피드

binwidth=60
df['first_work_avg_time'].sort_values().plot(kind='hist', bins=np.arange(min(df['first_work_avg_time']), max(df['first_work_avg_time']) + binwidth, binwidth))
work_avg_speed = df['first_work_avg_time'].median()
print('작업 평균 스피드: ' + str(int(work_avg_speed/60)) + '분 ' + str(int(work_avg_speed%60)) + '초')
plt.savefig('work_speed.jpg')

작업 평균 스피드: 2분 8초

오브젝트 수 / 데이터

#데이터 과부하 문제가 있을수있어 정식님에게 쿼리 요청하는게 나을듯

3. 설계 난이도 / 문제

작업자별 작업 수

df.first_work_cnt.value_counts().sort_index().plot(kind='bar', title='Number of Tasks Completed by Worker Distribution')
plt.xlabel('Number of Tasks Completed')
plt.ylabel('Worker Count')
plt.savefig('num_task_completed.jpg')

작업자별 재작업률

df['re_work_rate'] = df['re_work_cnt']/df['reject_cnt']
df['re_work_rate'].dropna().value_counts().sort_index().plot(kind='bar', title='Rework Rate by Worker Distribution')
plt.xlabel('Rework Rate')
plt.ylabel('Worker Count')
plt.savefig('rework_rate.jpg')

전체 재작업률

total_re_work_rate = df['re_work_cnt'].sum()/df['reject_cnt'].sum()
total_re_work_rate = 'Total Rework Rate: ' + str(total_re_work_rate)
print(total_re_work_rate)

Total Rework Rate: 0.9

반려 횟수 분포도

df3['check_reject_cnt'].value_counts().sort_index().plot(kind='bar', title='Reject Count Distribution')
plt.xlabel('Number of Rejections')
plt.ylabel('Data Count')
plt.savefig('check_reject_distribution.jpg')

이탈 작업자 Tracker

#n개의 작업을 한 작업자들을 필터 하고, 그들의 worker_id를 tuple data type으로 저장한다
worker_id = tuple(df[df['first_work_cnt'] <= params['n']]['worker_id'].values)

#tuple안에 있는 worker_id 값들을 string으로 전환시킨다
worker_id = tuple([[str(x) for x in worker_id] for y in worker_id][0])

#start_date와 end_date를 상황에 따라 input 값으로 바꿔야함
query = """
        SELECT worker_id, project_id, dt 
            FROM `crowdworks-platform.crowdworks_event.cw_event_data` 
            WHERE day BETWEEN "2021-06-03" AND "2021-06-10"
            AND worker_id IN {worker_id}
            AND action IN ('work_end')
    """.format(worker_id=worker_id)

raw_df = pd.read_gbq(query = query, project_id = 'crowdworks-platform', dialect='standard')

tracker_df = raw_df.groupby('worker_id')['project_id'].apply(set)
tracker_df = pd.DataFrame(tracker_df)

tracker_df.head()

	project_id
worker_id
105542	{7852, 7998, 7972, 7866}
110540	{7902}
112821	{8000}
118711	{8000, 7889, 7807}
12328	{7902}

counter = {}
for index, row in tracker_df.iterrows():
    for project_id in row['project_id']:
        if project_id not in counter:
            counter[project_id] = 0
        counter[project_id] += 1

# Data to plot
labels = []
sizes = []

for x, y in counter.items():
    labels.append(x)
    sizes.append(y)

plt.pie(sizes, labels=labels, autopct='%.2f')

plt.axis('equal')
plt.savefig('worker_tracker_pie.jpg')

리포트

project_name = "Project Name: " + project_info['project_name'][0]

data_type = "Data Type:  " + project_info['source_type_cd'][0]
project_type = "Project Type:  " + project_info['template_type'][0]
reward_worker = "Worker Unit Price:  " + str(project_info2['reward_worker'][0])
reward_checker = "Checker Unit Price:  " + str(project_info2['reward_checker'][0])
task_limit = "Task Data Limit:  " + str(project_info2['work_cnt_limit'][0])
start_date = "Start Date:  " + project_info['start_date'][0].strftime('%m/%d/%Y')
end_date = "End Date:  " + project_info['end_date'][0].strftime('%m/%d/%Y')
worker_salary = "Average Worker Salary:  " + str((3600/work_avg_speed)*project_info2['reward_worker'][0])
total_data_cnt = "Total Data Count:  " + str(project_info['total_src_cnt_pointer'][0])
days_left = (project_info['end_date'][0] - project_info['start_date'][0]).days
target_daily_rate = 'Target Daily Rate:  ' + str(int(project_info['total_src_cnt_pointer'][0]/days_left))

row_num = df2[df2['action']=='work_end'].groupby('day').count().shape[0]
actual_daily_rate = 'Current Daily Rate:  ' + str(df2[df2['action']=='work_end'].groupby('day').count()['dt'][row_num-2])

pdf = FPDF()
pdf.add_page()
pdf.set_xy(80, 5)

#제목
pdf.set_font('arial', 'B', 16)
pdf.cell(75, 10, "Project Analysis", 0, 2, 'L')

#1.Overview
pdf.set_xy(20, 20)
pdf.set_font('arial', 'B', 16)
pdf.cell(75, 10, "1. Overview", 0, 2, 'B')

#프로젝트 기본 정보
pdf.set_font('arial', 'B', 10)
pdf.set_xy(20, 35)
pdf.cell(0, 5, "Project Info", 0, 2, 'B')
pdf.set_font('arial', '', 10)
pdf.cell(0, 5, "Project Name: ", 0, 2, 'B')
pdf.cell(0, 5, project_type, 0, 2, 'B')
pdf.cell(0, 5, data_type, 0, 2, 'B')
pdf.cell(0, 5, total_data_cnt, 0, 2, 'B')
pdf.cell(0, 5, reward_worker, 0, 2, 'B')
pdf.cell(0, 5, reward_checker, 0, 2, 'B')
pdf.cell(0, 5, task_limit, 0, 2, 'B')
pdf.cell(0, 5, start_date, 0, 2, 'B')
pdf.cell(0, 5, end_date, 0, 0, 'B')

#작업 완료 스피드
pdf.set_xy(85, 35)
pdf.set_font('arial', 'B', 10)
pdf.cell(0, 5, "Work Rate", 0, 2, 'B')
pdf.set_font('arial', '', 10)
pdf.cell(0, 5, target_daily_rate, 0, 2, 'B')
pdf.cell(0, 5, actual_daily_rate, 0, 2, 'B')

#포인트
pdf.set_xy(85, 55)
pdf.set_font('arial', 'B', 10)
pdf.cell(0, 5, "Point", 0, 2, 'B')
pdf.set_font('arial', '', 10)
pdf.cell(0, 5, "Total Points: ", 0, 2, 'B')
pdf.cell(0, 5, worker_salary, 0, 2, 'B')

#진척률
pdf.set_xy(0, 90)
pdf.image('overview.jpg', x = None, y = None, w = 200, h = 0, type = '', link = '')

#Funnel
pdf.set_xy(40, 170)
pdf.image('funnel.jpg', x = None, y = None, w = 125, h = 0, type = '', link = '')

#1.작업 난이도
pdf.set_xy(20, 270)
pdf.set_font('arial', 'B', 16)
pdf.cell(75, 10, "2. Difficulty Level", 0, 2, 'B')

#작업 스피드
pdf.set_font('arial', '', 12)
average_speed = 'Average work speed: ' + str(int(work_avg_speed/60)) + ' min ' + str(int(work_avg_speed%60)) + ' sec'
pdf.cell(20, 15, txt=average_speed, border=0, ln=2, align='L')
pdf.set_xy(0, 30)
pdf.image('work_speed.jpg', x = None, y = None, w = 200, h = 0, type = '', link = '')

#평균 이미지당 오브젝트 수
pdf.set_font('arial', '', 12)
average_number_of_objects = 'Average Number of Objects per image: '
pdf.set_xy(20, 100)
pdf.cell(20, 15, txt=average_number_of_objects, border=0, ln=2, align='B')

#밴 rate
pdf.set_font('arial', '', 12)
ban_rate = 'Ban Rate: '
ban_count = 'Ban Count: '
pdf.set_xy(20, 120)
pdf.cell(0, 5, txt=ban_rate, border=0, ln=2, align='B')
pdf.cell(0, 5, txt=ban_count, border=0, ln=2, align='B')

#CS 수
pdf.set_font('arial', '', 12)
cs_rate = 'Number of CS per worker: '
cs_count = 'CS Count: '
pdf.set_xy(20, 140)
pdf.cell(0, 5, txt=cs_rate, border=0, ln=2, align='B')
pdf.cell(0, 5, txt=cs_count, border=0, ln=2, align='B')

#작업자 2
pdf.set_xy(20, 540)
pdf.set_font('arial', 'B', 16)
pdf.cell(0, 10, "3. Worker", 0, 2, 'B')

pdf.set_xy(0, 20)
pdf.image('active_worker_count.jpg', x = None, y = None, w = 200, h = 0, type = '', link = '')

pdf.set_xy(0, 90)
pdf.image('daily_average_task_count_by_worker.jpg', x = None, y = None, w = 200, h = 0, type = '', link = '')

#3.작업자
pdf.set_xy(20, 540)
pdf.set_font('arial', 'B', 16)
pdf.cell(0, 10, "3. Worker", 0, 2, 'B')

pdf.set_font('arial', '', 12)
pdf.set_xy(20, 25)
pdf.cell(0, 5, "Average Hours Worked per Day: ", 0, 2, 'B')
pdf.cell(0, 5, "Active Worker Number: ", 0, 2, 'B')

pdf.set_xy(0, 40)
pdf.image('num_task_completed.jpg', x = None, y = None, w = 200, h = 0, type = '', link = '')

#재작업률
pdf.set_xy(0, 110)
pdf.image('rework_rate.jpg', x = None, y = None, w = 200, h = 0, type = '', link = '')
pdf.set_xy(20, 180)
pdf.cell(0, 5, txt=total_re_work_rate, border=0, ln=2, align='B')

pdf.set_xy(0, 190)
pdf.image('check_reject_distribution.jpg', x = None, y = None, w = 200, h = 0, type = '', link = '')

#검수자
pdf.set_xy(20, 810)
pdf.set_font('arial', 'B', 16)
pdf.cell(0, 10, "4. Checker", 0, 2, 'B')

#플랫폼
pdf.set_xy(20, 1090)
pdf.set_font('arial', 'B', 16)
pdf.cell(0, 10, "5. Platform", 0, 2, 'B')

pdf.set_font('arial', '', 12)
pdf.set_xy(20, 25)
pdf.cell(0, 10, txt='Salary Comparison', border=0, ln=2, align='B')

pdf.set_xy(20, 40)
pdf.cell(0, 10, txt='Worker Exit Tracker', border=0, ln=2, align='B')
pdf.set_xy(0, 50)
pdf.image('worker_tracker_pie.jpg', x = None, y = None, w = 200, h = 0, type = '', link = '')

pdf.output('testing.pdf', 'F')

''